Sei sulla pagina 1di 10

International Journal of Heat and Mass Transfer 46 (2003) 31793188

www.elsevier.com/locate/ijhmt

A molecular dynamics simulation of droplet evaporation


Lorenzo Consolini 1, Suresh K. Aggarwal *, Sohail Murad
Department of Mechanical and Industrial Engineering, University of Illinois at Chicago, Chicago, IL 60607, USA
Received 27 August 2002; received in revised form 7 February 2003

Abstract
A molecular dynamics (MD) simulation method is developed to study the evaporation of submicron droplets in a
gaseous surrounding. A new methodology is proposed to specify initial conditions for the droplet and the ambient uid,
and to identify droplet shape during the vaporization process. The vaporization of xenon droplets in nitrogen ambient
under subcritical and supercritical conditions is examined. Both spherical and non-spherical droplets are considered.
The MD simulations are shown to be independent of the droplet and system sizes considered, although the observed
vaporization behavior exhibits some scatter, as expected. The MD results are used to examine the eects of ambient and
droplet properties on the vaporization characteristics of submicron droplets. For subcritical conditions, it is shown that
a spherical droplet maintains its sphericity, while an initially non-spherical droplet attains the spherical shape very early
in its lifetime, i.e., within 10% of the lifetime. For both spherical and non-spherical droplets, the subcritical vaporization, which is characterized by the migration of xenon particles that constitute the droplet to the ambient, exhibits
characteristics that are analogous to those reported for continuum-size droplets. The vaporization process consists of
an initial liquid-heating stage during which the vaporization rate is relatively low, followed by nearly constant liquidtemperature evaporation at a pseudo wet-bulb temperature. The rate of vaporization increases as the ambient
temperature and/or the initial droplet temperature are increased. For the supercritical case, the droplet does not return
to the spherical conguration, i.e., its sphericity deteriorates sharply, and its temperature increases continuously during
the vaporization process.
2003 Elsevier Science Ltd. All rights reserved.
Keywords: Molecular dynamics simulations; Submicron droplet; Subcritical and supercritical evaporation

1. Introduction
The physics of droplet vaporization has attracted
much interest during the past decades due to the crucial role played by sprays and atomizers in numerous propulsion, industrial, and environmental systems.
Amongst them, spray combustion perhaps represents
the most important application, considering the range of
devices that rely on liquidfuel combustion as their main
source of energy. Consequently, many researchers in this

Corresponding author. Tel.: +1-312-996-2235; fax: +1-312413-0447.


E-mail address: ska@uic.edu (S.K. Aggarwal).
1
Universit
a degli Studi di Roma, Tor Vergata, Rome,
Italy.

area have developed important fundamental theories


[1,2] on the evaporation of fuel droplets, taken as a
single entity or included in a spray. The droplet sizes
studied in these investigations have ranged from several
microns to a few millimeters. Modeling the vaporization
of such droplets has relied on the equations developed in
continuum uid mechanics. Although the sizes involved
were small, they remained large enough to neglect the
detailed molecular structure of the droplet. In addition,
the macroscopic sizes have allowed extensive experimentation on their evaporation behavior, thus facilitating validation of the continuum-based models.
Nonetheless, not much is known about the evaporation of droplets when their initial sizes are in the
submicron range. For this size range the continuum
assumption becomes questionable: the molecular structure of the droplet is no longer negligible, since the

0017-9310/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0017-9310(03)00101-7

3180

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

droplet size is only a few orders of magnitude greater


than the size of its constituent molecules. From a
mathematical perspective, many variables, such as
density, temperature, etc., which are assumed to be
continuous functions of space and time lose this characteristic. The proper description for the vaporization of
these smaller droplets, thus, requires a new methodology.
Molecular dynamics delivers this novel approach. A
molecular dynamics (MD) study of any physical phenomena attempts to simulate the real behavior of nature
by identifying each molecule and following their motion
in time through the basic laws of classical mechanics.
The system behavior and temporal evolution of its
thermodynamic and transport properties are, therefore,
nothing more than the cumulative result of individual
molecular motion.
The simulation of macroscopic droplet evaporation
using molecular dynamics is not yet possible due to the
prohibitively large computational requirements both in
terms of CPU time and memory. However, a system
involving a submicron droplet is small enough to successfully apply MD. Studying these smaller droplets
through MD will yield two primary benets: (1) such an
investigation will expand the knowledge of droplet
evaporation to sizes that are yet to be tested; (2) through
molecular analysis of the vaporization process, a better
understanding will be acquired of the molecular physics
of evaporating droplets of all sizes. Models developed in
the studies of larger droplets generally rely on certain
assumptions (for example: assuming the droplet to remain spherical throughout the process; neglecting second order eects such as the Soret and Dufour eects;
assuming thermodynamic equilibrium at the droplet
surface). A molecular model of a vaporizing droplet has
no need for these assumptions. Rather, if these were
correct, MD would simply conrm them.
Another motivation for developing MD approaches
to study droplet phenomena stems from the consideration that many important droplet processes involve
scales that cannot be resolved using continuum-based
methods. For instance, the phase change and interfacial
transport processes associated with transcritical and
supercritical droplet vaporization involve subcontinuum
scales, and, therefore, require phenomenological models
[14] for their representation in a continuum approach.
While such models may be appropriate to simulate
droplet vaporization under subcritical conditions, they
become questionable under transcritical/supercritical
conditions. The MD simulations can be used to examine
the applicability of these models under various conditions. The droplet vaporization under supercritical
conditions also involves signicant shape deformations
[5] due to vanishing surface tension. Since these deformations have length scales ranging from the droplet
size to the subcontinuum scale, continuum-based models

[68] may not be able to represent them accurately.


Moreover, atomization under supercritical conditions
would result in non-spherical droplets with a dynamically deforming interface. MD methods may be ideally
suited to simulate the vaporization and deformation
behavior of these droplets.
Motivated by these considerations, the present study
employs a MD approach to investigate the vaporization
of submicron droplets. The major objectives of this investigation are: (i) examine the eect of system size on
the simulated vaporization behavior, (ii) characterize
dierences and similarities between the vaporization of
subcontinuum and continuum-size droplets, (iii) characterize the eects of ambient and droplet properties on
the vaporization of submicron droplets, and (iv) examine the tendency of a droplet to remain spherical under
subcritical and supercritical conditions.
Previous investigations using MD methods for the
simulation of droplet vaporization have been relatively
few. Long et al. [9] used a MD methodology to simulate
the vaporization of an argon droplet in its own vapor
under subcritical conditions. Kaltz et al. [10] extended
this investigation to oxygen droplets evaporating in supercritical hydrogen and helium environments. The oxygen droplet was shown to exhibit a heating period
followed by a constant-temperature evaporation for
subcritical and moderately supercritical pressures. For
highly supercritical pressures, however, the droplet
temperature was shown to never reach a wet-bulb state,
increasing throughout the droplet lifetime. Walther and
Koumoutsakos [11] combined certain aspects of previous studies for the MD simulations of an argon droplet
in its own vapor. An adaptive tree data structure was
proposed to determine the list of neighbors surrounding
each particle, in order to avoid the computation of interatomic separation distance of one atom relative to all
the others.
The above investigations represent an excellent start
towards the development of MD methods for predicting
droplet vaporization under various conditions. However, considerable future eorts will be needed to make
MD methods a reliable simulation tool, and to apply
them to situations, including supercritical and deforming interfaces, that are not tractable using continuum
approaches. The future eorts should also focus on
simulating microscopic (of the order of microns) rather
than nano-droplets being investigated here, as well as
developing realistic yet tractable potential models for
more complex molecules to represent practical fuels. The
present study represents one such eort. It focuses on
the molecular physics of submicron droplet evaporation
for a new binary system: a xenon droplet evaporating in
nitrogen ambient. An alternative technique to identify
the droplet shape, as well as a dierent methodology for
specifying the desired initial conditions for droplet and
ambient are presented. In addition to examining the

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

eects of various parameters on the vaporization of


submicron droplets, the vaporization and shape histories
of initially non-spherical droplets under subcritical and
supercritical conditions are investigated.

2. Model formulation
The rst step in a MD simulation is to choose the
model that best represents the interaction between elementary particles. From a classical mechanics standpoint, it is commonly accepted that atoms and molecules
exert intermolecular forces, which are described through
a potential energy function. The present study employs
the Lennard-Jones 12-6 potential [12], whose analytical
form is
" 
 6 #
12
rij
rij
Uij rij 4eij 

1
rij
rij
In the above expression rij represents the intermolecular
separation distance between particles i and j. Furthermore, eij and rij are the minimum energy and the zero
energy separation distance relative to the pair. Data for e
and r of a wide range of atoms and small molecules are
widely available [13,14]. For cross interactions between
dierent species, it is common to use mixing rules such
as the Lorentz-Berthelot [15], which are given by
p
eij ei ej
2
rij 12  ri rj

The Lennard-Jones potential model used here is known


to be adequate for many non-quantum monatomic uids (Ar, Xe, Ne, etc.), as well as small diatomic and
polyatomic compounds (N2 , O2 , CH4 , etc.) [1316]. In
the present study, we have chosen to examine the
evaporation characteristics of Xe droplet in a N2 ambient. We chose the Xe/N2 system because this system
allows droplet behavior to be studied over a wide range
of conditions, including subcritical, transcritical, and
supercritical pressures. For this system, the following LJ
parameters were used: eXe 221kB , rXe 0:410 nm and
eN2 95:9kB , rN2 0:371 nm, where kB is the Boltzmanns constant. The cut-o radius was chosen to be
3rXe . The critical temperatures of N2 and Xe are 126.2
and 289.2 K respectively while the critical pressures are
33.9 and 58.4 bar respectively. This allows a rather wide
temperature region for a stable liquid droplet, with no
substantial condensation of the ambient gas into the
droplet.
In a MD simulation, specifying the total number of
particles N and the initial average number density d for
the simulated system yields the overall system volume:
V N =d, which in the present investigation is a cube.

3181

Initially all N particles are placed at the lattice sites of a


faced-centered cube (FCC) in the simulation system. The
liquid xenon droplet of an initial diameter D is placed at
the center of the cube, surrounded by the gaseous species
(nitrogen). A specied number of nitrogen molecules are
then removed from their FCC nodes in the ambient
(outside the droplet region) in order to establish the
desired initial ambient density. For example, in a typical
system studied, N 5662, while the number of molecules constituting the droplet and the ambient are 3559
and 2103, respectively. It should be further noted that
the volume occupied by the ambient is considerably
larger than that occupied by the droplet in our simulations (cf. Fig. 1).
The numerical scheme adopted to solve the Newtonian equations of motion for every particle is based
on the Gears fth order predictorcorrector algorithm
[12]. The nature of the Lennard-Jones interaction is such
that the acceleration of any one molecule, at a given instant
in time, is only a function of the separation distances, rij .
Therefore, setting the initial positions corresponds to
setting the initial accelerations. Gears algorithm also
requires the assignment of initial velocities of all the
system particles. These are assigned randomly as a
Gaussian distribution based on the desired initial temperature, using the equations of statistical mechanics
[12].
The conditions for the numerical algorithm to initiate
the computation of particle trajectories are now specied. As the particles start to move, the xenon crystal
melts, losing memory of its initial orderly structure and
acquiring the structure of the liquid state. As the simulation evolves, some xenon atoms evaporate and mix
with nitrogen molecules and the vapor content of the
ambient increases. Due to the small sizes involved (the
initial droplet diameter is taken to be 6.88 nm), distinguishing the condensed phase from the gaseous phase
becomes an important challenge. The identication of
the droplet is based on dynamically monitoring the
proles of liquid particle number density in the three
system coordinates. Assuming the droplet to be inscribed in a regular ellipsoid, the length of its x-axis is
calculated using the following procedure. Imagine slicing the system cube into forty equal slices in the x direction, with each slice housing a certain number of
particles. The length of the x-axis of the droplet is then
calculated according to the number of continuous slices
with xenon particles greater than (or equal to) a given
threshold value based on the liquid density. Additional
details are provided in Ref. [17].
The time interval required by the xenon FCC crystal
to become a liquid droplet will be referred to as the
initialization period. During this period the crystal is left
to melt and no datum is collected. Also, during this
period, the velocities of both xenon and nitrogen particles are rst scaled to achieve the desired initial droplet

3182

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

Fig. 1. Snapshots of the xenon droplet during the vaporization process at (a) t 0, (b) 510 ps, (c) 926 ps and (d) 1030 ps. Only xenon
molecules are shown.

and ambient temperatures, respectively, while maintaining a motionless center of mass. Details are provided
in Ref. [17]. Once the velocity scaling is stopped, which
typically involves 100 time steps, the system is left to
adjust further for a brief time period (typically 1000 time
steps). During this period, the ambient and droplet
particle velocities will generally adapt to the system
conguration. The end of the initialization period corresponds to the start of the production part of the
simulation and, therefore, represents the origin of time.

density of 2600 kg/m3 (2000 xenon atoms), evaporating


in a subcritical nitrogen environment. The droplet conguration is shown at dierent instances in its lifetime,
with its liquidvapor interface identied using the procedure outlined earlier. A crown of vapor atoms appears to be consistently surrounding the droplet. In
addition, the vaporizing droplet is shown to lose, to
some extent, its spherical shape at random instances in
its lifetime. An important observation here is that the
algorithm developed to distinguish droplet from ambient does an excellent job, and, to a large extent, the
droplet maintains its spherical shape.

3. Results
3.2. Eect of initial droplet and system sizes
3.1. Droplet evaporation
In order to assess the capability of the MD model to
distinguish droplet from ambient during the vaporization process, snapshots of the simulated system were
recorded at dierent instances in time. Fig. 1 presents
these snapshots or instantaneous pictures for a xenon
droplet, with an initial diameter of 6.88 nm and initial

To assess the sensitivity of the MD simulations to the


initial droplet and system size used, results were obtained for three systems of dierent sizes. The droplet
diameters for the three systems were 3.44, 5.14, and 6.88
nm, and the corresponding number of xenon particles
were 205, 783, and 2000, respectively. For all three cases,
the initial ambient (nitrogen) temperature and density

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

were 650 K and 8.5 kg/m3 respectively. The volumes of


the primary system domain were 2.6  103 nm3 8.8  103
nm3 and 20.8  103 nm3 respectively.
Fig. 2 presents the evaporation behavior for the three
cases. The number of droplet particles, normalized by
the initial value, is plotted versus non-dimensional time,
s, which is dened as the actual time elapsed, t, divided
by the time required for half the droplet to evaporate,
t1=2 : s t=t12 . Half the droplet is considered evaporated
when half the initial droplet particles have migrated to
the vapor phase. The simulations are essentially inde-

3183

pendent of the system size, although uctuations are


larger for the smaller system.
The droplet sphericity is a good measure of the tendency of a droplet to maintain a spherical shape during
the vaporization process. The droplet sphericity, S, is
dened as the ratio of the smallest half axis length to the
largest:
S

minfa; b; cg
maxfa; b; cg

Here a, b, c represent the three droplet axial lengths. Fig.


3 shows the temporal variation of S during the vaporization process for the three system sizes. As the system
size increases, the sphericity appears to maintain a value
consistently closer to unity, implying that under subcritical conditions, a submicron droplet maintains the
spherical shape during its lifetime.
3.3. Vaporization history

Fig. 2. Total number of droplet particles, normalized by their


initial value, plotted versus the normalized time for the three
systems of dierent initial sizes.

To examine the fundamental physics of submicron


droplet evaporation, the behavior of a xenon droplet
was analyzed for dierent ambient conditions. The initial ambient density was 8.5 kg/m3 and the initial pressures for the three cases were 26, 16 and 9 bar,
corresponding to the initial ambient temperatures of
1030, 650 and 330 K respectively (after the initialization
period). The initial droplet temperature was xed at 165
K. It should be noted that the system pressure changes
during the simulation due to two eects. The rst is due
to the decrease in ambient temperature, caused by the
heat transfer from ambient to droplet and by the addition of cooler xenon vapor molecules into the ambient
(Fig. 5b). This decreases the system pressure. On the

Large System

Medium System

Small System
Fig. 3. Droplet sphericity plotted versus the normalized time for the three systems discussed in the context of Fig. 2.

3184

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

Fig. 4. Temporal evolution of the number of droplet particles for dierent ambient temperatures.

other hand, the system pressure increases as the ambient


density increases due to the addition of xenon vapor
molecules. Since the eect of ambient temperature generally dominates, the system pressure decreases during
the simulation. For example, for the case with the ambient temperature of 330 K, the system pressure decreases from 9 to 7 bar during the simulation time of
500 ps.
Fig. 4 presents the vaporization history in terms of
the temporal variation of the number of droplet particles
for three dierent ambient temperatures. The corresponding droplet and ambient temperature histories are
shown in Fig. 5. As expected, the evaporation process is
characterized by the loss of liquid xenon particles from
the droplet, and their migration to the ambient. The
vaporization rate increases as the ambient temperature is
increased; for example, increasing the ambient temperature from 330 to 1030 K increases the vaporization rate
by a factor of more than three. The enhanced vaporization rate is directly attributable to the higher droplet
temperature corresponding to the higher ambient temperature. The droplet temperature plots, presented in
Fig. 5a, exhibit two distinct stages of droplet vaporization: (1) a liquid-heating stage characterized by
relatively slow vaporization, and (2) a constant-liquidtemperature evaporation at the pseudo wet-bulb temperature. It is interesting to note that a similar behavior
is observed for larger or continuum-size droplets
evaporating under subcritical conditions [1,2]. In addition, Fig. 5a indicates that both the liquid heat-up time
and the pseudo wet-bulb temperature increase as the
ambient temperature is increased. This behavior is also
analogous to that observed in continuum-based studies
dealing with larger droplets. It should also be noted that
for the lowest ambient temperature case, the droplet

temperature decreases with time. This is attributed to the


fact that due to the relatively small temperature dierence between the droplet and the ambient gas, the heat
supplied from the ambient gas is less than that required
for vaporization. A similar behavior has been observed
in studies dealing with continuum-size droplets [18].
In Fig. 5b, we present the temporal variation of
ambient temperature during the evaporation process.
The ambient temperature in the context of MD simulations represents an average temperature of all the nitrogen and xenon particles that are in the ambient. The
ambient temperature decreases with time due to the incoming cooler xenon vapor particles, and due to the
transfer of energy from ambient to droplet. The behavior is similar to that observed for macro-size droplets. For the latter case, the gas lm near the droplet
experiences a similar decrease in temperature, and the
gas-lm thickness is of the order of the droplet diameter.
For the MD simulations, the ambient size is of the order
of the submicron droplet diameter, and its temperature
decreases due to heat transfer to the evaporating droplet, and due to the addition of xenon vapor particles into
the ambient.
3.4. Evaporation of an initially non-spherical droplet
One of the fundamental assumptions in the study of
macroscopic droplet evaporation has been that the
droplet remains spherical during its lifetime. For subcritical vaporization, especially under non-convective or
mildly convective conditions, the assumption is justied
since the Weber number is relatively small due to the
nite surface tension. However, under transcritical and
supercritical conditions, as surface tension goes to zero,
the assumption of spherical shape becomes highly

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

3185

220

0.9

200

0.8
0.7

180

0.6

160

0.5
0.4

140

(a)

0.3
0.2

120

(a)

22

100
0

85

170

255

340

425

65

87

500

130

(b)

450

1250

108

Time (picoseconds)

510

Time (picoseconds)

43

400
350

1000

300
250

750

200
150

500

100
0

250

29

58

86

115

144

173

Time (picoseconds)

(b)

Subcritical

Supercritical

0
0

85

170

255

340

425

510

Time (picoseconds)

Amb.Temp.:1030K

Fig. 6. Temporal variation of (a) sphericity and (b) temperature of an initially elliptical droplet evaporating under subcritical and supercritical conditions.

Amb. Temp.: 650K

Amb.Temp.:330K
Fig. 5. Temporal evolution of (a) droplet temperature, and (b)
ambient temperature for the three cases discussed in the context
of Fig. 4.

questionable. In order to address this issue, MD simulations are used to study the vaporization of initially
non-spherical droplets under both subcritical and supercritical conditions. An initially elliptic xenon droplet
with an initial density of 2250 kg/m3 and comprising of
1800 atoms is considered. For the subcritical case, the
initial liquid temperature is 165 K, while the ambient
temperature and pressure are 650 K and 16 bar respectively. For the supercritical case, the initial liquid temperature is 266 K, and the ambient temperature and
pressure are 1320 K and 78 bar respectively.
Fig. 6 presents the temporal variation of sphericity
(S) and temperature for an initially elliptic droplet for
both the subcritical and supercritical conditions. For
both cases, the initial value of S is 0.65, dened by the
following half-axis values: a 4:50 nm, b 2:90 nm
and c 3:25 nm. For the subcritical case, the initially
elliptical droplet attains the spherical shape rather early

in its lifetime, and its vaporization follows the typical


subcritical behavior, which has been described in the
preceding section. The droplet temperature increases to
the quasi wet-bulb temperature value, as indicated in
Fig. 6b, with the subsequent vaporization occurring at
this temperature. The rapid attainment of the spherical
shape for an initially elliptical droplet under subcritical
conditions is also shown by the snapshots of xenon
atoms in Fig. 7. In contrast, for the supercritical case, the
droplet never returns to the spherical, i.e., the minimum
energy conguration. Rather, sphericity of the droplet
deteriorates rapidly and noticeably, and its temperature
continues to increase throughout the vaporization
process (cf. Fig. 6). Similar variation in droplet temperature under supercritical conditions has been observed in the study of macroscopic droplets [6,19], as
well as that of nano-size droplets [10].
As discussed in the context of Fig. 6, a subcritical
microscopic droplet, with a disrupted initial shape,
has a strong tendency to recover the spherical conguration. The driving force that leads the droplet back to its
traditional shape is determined primarily by the liquid
temperature. To evaluate the eect of liquid temperature

3186

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

Fig. 7. Snapshots of xenon atoms for an initially non-spherical droplet at (a) t 0 ps, (b) 10 ps, (c) 16 ps and (d) 36 ps for the
subcritical case. Only xenon particles are shown.

1
Dr oplet Spher icity, S

on the rate of change of droplet sphericity, the behavior


of the subcritical droplet was studied under dierent
initial conditions.
The inuence of initial droplet temperature on the
temporal evolution of droplet shape is presented in Fig.
8. The three droplets become nearly spherical within few
picoseconds. In addition, it is observed that the tendency
of a non-spherical droplet to attain a spherical shape is
reduced as the initial droplet temperature is increased.
For example, at t P 50 pico-seconds (ps), the droplet
with higher initial temperature (T 195 K) is less
spherical than those with lower initial temperatures
(T 165 and 130 K). Surface tension is a measure of the
overall attractive force acting on the surface, exerted by
the inner bulk molecules, that causes the droplet surface
to attain the spherical shape. Therefore, in an equilibrium quiescent environment, a droplet with a higher
surface tension will have a greater tendency to remain
spherical. Since surface tension generally decreases with
increasing liquid temperature, the results in Fig. 8 are

0.9
0.8
0.7
0.6
0.5
0.4
0

26

52

78

104

130

156

Time (picoseconds)
Dr oplet Temp.: 195K
Dr oplet Temp.: 130K

Dr oplet Temp.: 165K

Fig. 8. Temporal variation of the sphericity of three initially


elliptical droplets with dierent initial temperatures.

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

3187

consistent with this behavior. However, for t < 25 ps,


the results in Fig. 8 may appear to be somewhat in
contrast with this behavior, but they simply represent
uctuations in the droplet shape, and the amplitude of
uctuations generally increases at higher liquid temperature as observed in our simulations.

higher initial temperature becomes less spherical. For


the supercritical case, the droplet never returns to the
spherical conguration. Rather, sphericity of the droplet
deteriorates rapidly and noticeably, and its temperature
increases continuously during the vaporization process.

4. Conclusions

Acknowledgements

Molecular dynamics (MD) simulations have been


used to study the vaporization behavior of submicron
droplets. A new methodology has been developed to
specify initial conditions for the droplet and ambient,
and to identify droplet shape during the vaporization
process. The vaporization characteristics of a xenon
droplet in nitrogen ambient have been investigated.
Results have focused on (1) the sensitivity of MD simulations to the system and droplet sizes, (2) the vaporization behavior of spherical and non-spherical droplets,
(3) the eects of ambient and droplet properties on the
vaporization process, and (4) the tendency of a droplet
to retain its spherical shape under both subcritical and
supercritical conditions.
The MD simulations are shown to be essentially independent of the primary system size, although the
predicted vaporization behavior shows less scatter when
a larger system is employed. The predicted droplet
sphericity is also shown to be higher, i.e., closer to unity
with the larger system.
The evaporation process is characterized by the migration of xenon particles from the droplet to the ambient. The rate of migration or vaporization increases as
the ambient temperature is increased, which is directly
attributable to the higher droplet temperature corresponding to the higher ambient temperature. Analogous
to the vaporization of macro-size droplets, the vaporization process associated with submicron droplets
consists of a liquid-heating stage during which the vaporization rate is relatively low, followed by a nearly
constant liquid-temperature evaporation at the pseudo
wet-bulb temperature.
MD simulations are also used to examine the tendency of a submicron droplet to retain its spherical
shape under subcritical and supercritical conditions. The
vaporization of an initially elliptic droplet is analyzed.
For the subcritical case, an initially elliptical droplet
attains the spherical shape rather early in its lifetime,
and its vaporization follows the typical subcritical behavior. The tendency to return to spherical shape is
initially (at t 6 25 ps) higher for a droplet with higher
initial temperature. This can be attributed to the highly
non-equilibrium process initially and the enhanced loss
of the higher energy vertex molecules, even though the
surface tension is reduced at higher liquid temperature.
At later times (t P 50 ps), however, the droplet with

This research was supported in part by a grant from


the Chemical Sciences Division, US Department of
Energy (DE-FG02-96ER14680). The support from the
Campus Research Board at UIC is also gratefully acknowledged.

References
[1] W.A. Sirignano, Fluid Dynamics and Transport of Sprays,
Cambridge University Press, UK, 1999.
[2] S.K. Aggarwal, A. Tong, W.A. Sirignano, A comparison
of vaporization models for spray calculations, AIAA J. 22
(10) (1984) 14481457.
[3] K.C. Hsieh, J.S. Shuen, V. Yang, Droplet vaporization in
high pressure environments I: near critical conditions,
Combust. Sci. Tech. 76 (1991) 111132.
[4] G. Zhu, S.K. Aggarwal, Transient supercritical droplet
vaporization with emphasis on the eects of equation of
state, International Journal of Heat and Mass Transfer 43
(2000) 11571171.
[5] B. Chehroudi, Private Communication, 2001.
[6] S.D. Givler, J. Abraham, Supercritical droplet vaporization and combustion studies, Prog. Energy Combust. Sci.
22 (1996) 128.
[7] J.P. Delplanque, W.A. Sirignano, Numerical study of the
transient vaporization of an oxygen droplet at sub- and
super-critical conditions, Int. J. Heat Mass Transfer 36
(1993) 303314.
[8] V. Yang, Modeling of supercritical vaporization, mixing,
and combustion processes in liquid-fueled propulsion
systems, Proc. Combust. Inst. 28 (2000) 16441662.
[9] L.N. Long, M.M. Micci, B.C. Wong, Molecular dynamics
simulation of droplet evaporation, Comput. Phys. Commun. 96 (1996) 167172.
[10] T.L. Kaltz, L.N. Long, M.M. Micci, J.K. Little, Supercritical vaporization of liquid oxygen droplets using
molecular simulations, Combust. Sci. Tech. 136 (1998)
279301.
[11] J.H. Walther, P.J. Koumoutsakos, Molecular dynamics
simulation of nanodroplet evaporation, J. Heat Transfer
123 (2001) 741748.
[12] J.M. Haile, Molecular Dynamics, Elementary Methods,
Wiley, New York, 1992.
[13] J.O. Hirschfelder, C.F. Curtiss, R.B. Bird, Molecular
Theory of Gases and Liquids, fourth ed., John Wiley &
Sons, New York, 1967.
[14] L.L. Lee, Molecular Thermodynamics Nonideal Fluids,
Butterworths, Boston, 1988.

3188

L. Consolini et al. / International Journal of Heat and Mass Transfer 46 (2003) 31793188

[15] M.P. Allen, D.J. Tildesley, Computer Simulation of


Liquids, Oxford University Press, 1984.
[16] T.M. Reed, K.E. Gubbins, Applied Statistical Mechanics,
McGraw Hill, New York, 1973.
[17] L. Consolini, A molecular dynamics simulation of droplet
evaporation, MS Thesis, University of Illinois at Chicago,
IL, 2002.

[18] S.K. Aggarwal, G. Chen, T.A. Jackson, G.L. Switzer,


Vaporization behavior of pure and multicomponent fuel
droplets in a hot air stream, Int. J. Heat Mass Transfer 34
(10) (1991) 26692673.
[19] H. Jia, G. Gogos, High pressure droplet vaporization;
eects of liquid-phase gas solubility, Int. J. Heat Mass
Transfer 36 (1993) 44194431.

Potrebbero piacerti anche